A nonlinear, bistable optical device, referred to as a self electro-optic effect device (SEED) and described in U.S. Pat. No. 4,546,244 issued to D. A. B. Miller on Oct. 8, 1985, has a semiconductor quantum well region that is electrically controlled to change the state of the device via a photodetector. Many SEED devices must be interconnected in typical signal processing applications such as optical computing or photonic switching. Two approaches have been proposed for interconnecting the devices. In one approach, the devices are constructed in a manner where optical signals travel through the devices in optical waveguides. Waveguides are also used to interconnect several devices on a chip in a manner similar to that used for titanium-diffused lithium niobate directional couplers. Devices on different chips are then interconnected with optical fibers. A disadvantage of this approach is that not many devices can be integrated on a single substrate because of the amount of room required for the waveguide interconnections.
A second approach is to build the devices in two-dimensional arrays, where the optical signals travel normal to the plane of the devices. The devices are optically interconnected using bulk optical elements such as lenses and beam splitters. However, the bulk element arrangements required to perform even relatively simple optical interconnections may be quite complex.
In view of the foregoing, a need exists in the art for a simple optical interconnection mechanism for interconnecting a large number of optical devices.